Differential privacy

Interesting article from Apple on privacy: here

Understanding how people use their devices often helps in improving the user experience. However, accessing the data that provides such insights — for example, what users type on their keyboards and the websites they visit — can compromise user privacy. We develop a system architecture that enables learning at scale by leveraging local differential privacy, combined with existing privacy best practices. We design efficient and scalable local differentially private algorithms and provide rigorous analyses to demonstrate the tradeoffs among utility, privacy, server computation, and device bandwidth. Understanding the balance among these factors leads us to a successful practical deployment using local differential privacy.



cookieOK, so cookies are actually only mentioned one time in GDPR, but that one time packs a bit of a punch.

Natural persons may be associated with online identifiers… such as internet protocol addresses, cookie identifiers or other identifiers… This may leave traces which, in particular when combined with unique identifiers and other information received by the servers, may be used to create profiles of the natural persons and identify them.

Which translated basically means (taken with other parts of the regulation) if you can identify an individual via their device (directly or indirectly) that makes it personal data.

Now not all cookies will be able to identify users but a whole load of them are. And that includes analytics cookies.

The existing “cookie law” was pretty clear about gaining consent and we all added those cookie bars on to our websites that basically implicitly gained your permission. Well those aren’t any good any more. For any cookies that aren’t strictly necessary to run your site you’re going to have to get explicit consent under GDPR.

That means you need some kind of affirmative action. Like a tick box.

And that consent must be as easy to take away as it was given.

Oh, and it’s about consent. So if you’re not giving a choice then there’s no consent to be given or not. Which basically means you can’t tell your visitors they have to accept or the can’t browse…

Insight at the heart – changing culture

img_0032I have a mantra. It goes like this: you cannot call yourself customer centric unless you are first data centric.

In other words, your data (and the insight you service from it) will tell you so much about your customers that if you’re not using it to the absolute maximum then you can’t be providing the best to your customers that you can.

But shifting the culture in an organisation is tough. You can stand up on your soapbox and suddenly proclaim that your business will make decisions based on data, but those are just words. How do you get everyone to follow?

For me the greatest successes I’ve had in getting big organisations to start being more data and insight focused is to start small. Put away that soapbox and instead start building a movement behind you. Get a couple of senior stakeholders bought into what you are trying to achieve. And then get a couple of your peers bought in too. And between you start to create a plan of action and sketch out your processes.

And then get on with it. Accept you won’t get everything right immediately. But because your movement is small it doesn’t matter. You haven’t gone out there and promised the Earth. You can learn and try again.

And once you have a couple of really good wins behind you, start talking about it. Again, don’t go out there with your big bang, but just build your movement out a little further and embrace a few more followers. Keep your senior stakeholders in the loop and excited about what you are doing, and continue to test and learn as your new followers come on board.

When you feel like you have a bit of momentum behind you and crucially you have strong processes and governance in place (good quality data, people, escalations, etc) then go ahead and talk a little more widely and a little more loudly. More people will jump on board. You’ll have a few more challenges and you will evolve but that’s only to the benefit of the movement.

This won’t all happen overnight. You will need to be in it for the long run. But you are more likely to have sustained success and growth this way than shouting at people to follow you (or in some places I’ve been in, to just fucking do it…).

Customers ‘bewildered and fearful’ about use of their data – BBC

Nine in 10 people have no idea what companies do with the personal information the firms hold about them, a survey suggests.

The Chartered Institute of Marketing (CIM) survey of 2,500 people also found 57% did not trust the companies to handle their data responsibly.
And 51% complained that they had been contacted by organisations that had misused their data.

Read more: BBC News – Customers ‘bewildered and fearful’ about use of their data

Google and automated insights

Robot armGoogle continues to add some interesting features to Google Analytics, and one of the latest is the ability to get automated insights.

It uses Google’s machine learning to comb through the data you have and comes up with a stream of what it calls automated insights.

For some reason they have decided to make this feature available in the Android and iOS Google Analytics app, but not the desktop interface (presumably they want more people to take the app right now…).

In the Google Analytics blog announcement which you can read here it specifically calls out “marketers, business owners, and product designers” but doesn’t mention analysts at all. 

I’m a bit fan of automation when and wherever it makes sense. But I’m still not yet convinced that machine learning is clever enough to properly interpret everything that’s going on and the danger here is that people without full knowledge make bad decisions because Google Analytics told them too.

A little bit of the cynic in me says a lot of the “insight” that Google Analytics users will be seeing will be telling people they need to be advertising more with Google…


helpThere is no right answer to creating a good online self-service environment, but there are some strong best practice pointers that will help guide the way. Not everything will be suitable for every business.

Here are some pointers though!

  • Ensure you have joined up web analytics and reporting, including contact centre, voice of the customer and web data.
  • Use simple classifications to help people filter quickly and easily.
  • Don’t have long lists of FAQs. Have a clear list of top questions, and then have a clear path for other users to filter to the right answer.
  • Make sure your search engine is capable of using natural language queries, and that it provides a small handful of relevant results. Implement a “did you mean” feature for clear filtering, and include either automatically corrected spellings or corrections.
  • Not all content has to be provided by you. Other sites might be better able to answer the question.
  • Have a user forum/community. Your engaged base will help out other customers in need of help. It’s also good for SEO, helps with appropriate escalation and is good for the brand. Ensure the community is fully integrated into your online help solution, including search.
  • Consider using video for your high traffic solutions. But also consider using your engaged base to help make those videos, and reward them for doing it.
  • Add a full customer satisfaction survey, and if possible include session replay.
  • Identify areas where contact is required, and ensure escalation points are clearly signposted. Finding an answer should be easy, even if that means the customer has to make a phone call rather than spend time searching the website. Consider web chat for help.
  • Ensure content is accessible on mobile.
  • Allow people to rate every answer on the website. This helps compliment web analytics with a partial voice of the customer. It also helps to weed out potentially poor answers for continual optimisation.

Ensure you use consistent language across your digital touchpoints, from purchase to account management to online help, store and the contact centre. Create a language bible that is shared across the business. For example, a talk plan should be referred to consistently across the brand.

Don’t work in silos. Some lessons from the US:

“A lack of coherence can damage brand image, and because of the confusion caused by the variance in navigation systems, could lead to frustration on the part of the customer and to increased calls to the telephone help desks.…support has a different look and feel and uses a different menu system. This can also slow the customer down when looking for help. Customer confusion poses a significant challenge, as visitors’ patience levels are low”[i].

Furthermore “…best practices demand that businesses be equipped to manage the customer experience via the preferred channel of the customer – whether it’s online via self-service, online via assisted service, or offline through a phone or in person. For some businesses, this process can be hindered by silos of informational hierarchies – with marketing owning the web site, contact centre owning many of the customer interactions and with neither communicating effectively with the other.”[ii]

[i] Verizon report, Customer Respect Group, 2010

[ii] IntelliResponse, Web Self-Service: The Cornerstone of Multi-Channel Customer  Experience Management